Trim convergence fails for above-rated operating points when linearizing the DTU 10 MW (OpenFAST 5.0.0)

[jefvar]

Hello,
I am setting up linearization of the DTU 10 MW RWT in OpenFAST v5.0.0 (model converted from an older FAST v8 / AeroDyn v14 input set) and I would appreciate some feedback on trim convergence.
My goal is to generate linearized models at 10 operating points from 5 to 25 m/s. For each case I use steady uniform wind and I change: the wind speed (InflowWind), the initial rotor speed RotSpeed (ElastoDyn), and TrimCase (2 below rated, 3 above rated), with CalcSteady = True. To validate the setup first, I enabled only GenDOF and checked whether the generator speed converges to the target before worrying about the full DOF set.
The first operating point (below rated) behaves reasonably, but for the above-rated cases the rotor decelerates from the start instead of holding the target speed, and I get:

The BEM solution is being turned off due to low TSR. (TSR = 1.9995).
This warning will not be repeated though the condition may persist. (See GeomPhi output channel.)

so the rotor ends up collapsing to a very low TSR rather than converging to the periodic steady state.
Linear models.zip

I attach the full input file set (.fst, ElastoDyn, AeroDyn + blade file, ServoDyn, InflowWind). Thanks in advance for any help.
PD: run calcSsmats.m first

[jjonkman]

Dear @jefvar,

I can foresee a couple problems. TrimGain likely needs to be smaller for TrimCase = 3, e.g. TrimGain = 0.1 rad/(rad/s) in the OpenFAST primary (.fst) input file. Also, we generally recommend when using TrimCase = 3 that the generator torque be held constant, e.g., set VS_RtGnSp = 1 rpm in ServoDyn. Does that help?

If not, what do the rotor speed, blade-pitch angle, and generator torque time series look like for an above rated case?

Best regards,